【作者】 张群洪；

【导师】 王钦敏；

【作者基本信息】 福州大学 ， 计算机应用技术， 2004， 硕士

【摘要】 随着数据库应用技术以及空间数据矢量化技术的迅速发展，人们积累的数据越来越多，大量的空间数据存储在空间数据库中。如何发现在大型空间数据库中所隐藏的、预先未知的信息以辅助相应的应用，这就是目前空间数据挖掘的任务。同时空间关联规则是空间数据挖掘结果的一种最主要的知识规则，它侧重于确定数据中不同领域之间的联系，找出满足给定支持度和可信度阈值的多个域之间的依赖关系。本文主要研究工作如下：1 阐述空间数据结构、空间数据库和空间数据挖掘的相关理论和概念，并介绍几种目前比较典型的空间数据挖掘原型系统。2 讨论了几种空间数据预处理方法：空缺数据填充、噪音数据处理、数据规约、连续属性的离散化以及基于面向属性归纳的空间特征概化。其中连续属性离散化包括常规离散化方法、基于Rough理论的GIS属性数据约简和基于云理论的属性数据泛化。3 研究关联规则Apriori算法，分析了传统的关联规则理论基础、经典算法，探讨了提高Apriori算法效率的几种方法，着重介绍一种不产生候选挖掘频繁项集的方法。在充分理解理论基础上，文中提出了一种改进的关联规则算法：一种基于映射技术的关联规则采掘算法MBAR，并进行简单的测试，验证算法的有效性。4 论文分析了空间关联规则挖掘的一般理论和方法以及空间关联规则挖掘对象的特点，并提出了一种基于概念树的多层次空间关联规则挖掘算法，算法分为两个部分，第一部分利用GIS中的空间分析技术和空间关系运算等对地理空间中的目标和对象的空间关系以及空间行为进行描述，生成以空间属性为主的空间概念层次关系；第二部分提出利用此空间概念层次关系进行空间关联规则挖掘的算法。这种算法有效地利用了GIS中的空间分析技术，较好地处理了空间数据间的空间关系，通过复杂的空间计算和分析求得空间属性之间的关联规则，大大提高了挖掘效率。在相关理论研究的基础上，设计了一个基于J2EE的空间数据挖掘原型系统。系统是以通用空间数据库为核心，包含三个功能模块：空间数据管理和人机交互模块；空间数据挖掘模块(Spatial Data Mining，SDM)；数据库服务模块。并以龙海市土地利用为例进行实例验证。更多还原

【Abstract】 With the development and application of database technology and Vector Map Technology，large quantities of spatial data have been produced and stored in spatial database and spatial data warehouse. The main task of spatial data mining is to discovery the implicit, previous unknown, and potential useful information from these data. In the same time, spatial association rules is one of the upmost knowledge rules in the result of spatial data mining. It emphasizes particularly on confirming the relation of data in different fields. It tries to find out the dependence of data in multi-fields. The remainder of this paper is organized as follows:1. In the paper we expound the general theories and concepts of the spatial data structure, spatial database and spatial data warehouse, and introduce several typical prototype systems of spatial data mining. 2. We discuss several methods of spatial data preprocessing: empty data filling, noise data processing, data reduction, spatial characteristic generalizaion based on attribute induction, and sequent data discretization, which includes common discretized methods, GIS attribute data reduction based on Rough theory, and attribute data generalization based on cloud theory.3. In the paper we research the Apriori arithmetic of association rules, probe into several efficient methods to improve the Apriori arithmetic, and introduce emphatically a mining method without generating candidate frequent itemsets: FP-tree. After understanding well the association rules, an improved Apriori arithmetic: MBAR((Map-based association rule mining),has been developed and tested in the paper.4. We analyze the general theories of spatial association rules, especially bring out a multiple-level spatial association rules arithmetic. The algorithm improves the efficiency greatly by complex spatial computing and analyzing to gain the association between spatial attributes, accordingly, it avoids the complex process of gaining frequent itemsets. On the base of the relative theories above, we design a spatial data mining prototype system based on J2EE, which includes three functional modules: spatial data management and interactive module based on Web-GIS,SDM(spatial data mining) module, and database service module. We take the land use of longhai city as an example, and test the system on it.更多还原